ARTICLEPEDIATRICS Volume 137 , number 6 , June 2016 :e 20151353

Neonatal Phototherapy and Infantile CancerAndrea C. Wickremasinghe, MD, a, b Michael W. Kuzniewicz, MD, MPH, c, d Barbara A. Grimes,

PhD, b Charles E. McCulloch, PhD, b Thomas B. Newman, MD, MPHb, c, d

abstractOBJECTIVE: To determine whether neonatal phototherapy is associated with cancer in the first year after birth.METHODS: We analyzed a data set from the California Office of Statewide Health Planning and Development that was created by linking birth certificates, death certificates, and hospital discharge abstracts up to age 1 year. Subjects were 5 144 849 infants born in California hospitals at ≥35 weeks’ gestation from 1998 to 2007. We used International Classification of Diseases, Ninth Revision codes to identify phototherapy at <15 days and discharge diagnoses of cancer at 61 to 365 days. We adjusted for potential confounding variables by using traditional and propensity-adjusted logistic regression models.RESULTS: Cancer was diagnosed in 58/178 017 infants with diagnosis codes for phototherapy and 1042/4 966 832 infants without such codes (32.6/100 000 vs 21.0/100 000; relative risk 1.6; 95% confidence interval [CI], 1.2–2.0, P = .002). In propensity-adjusted analyses, associations were seen between phototherapy and overall cancer (adjusted odds ratio [aOR] 1.4; 95% CI, 1.1–1.9), myeloid leukemia (aOR 2.6; 95% CI, 1.3–5.0), and kidney cancer (aOR 2.5; 95% CI, 1.2–5.1). The marginal propensity-adjusted absolute risk increase for cancer after phototherapy in the total population was 9.4/100 000 (number needed to harm of 10 638). Because of the higher baseline risk of cancer in infants with Down syndrome, the number needed to harm was 1285.CONCLUSIONS: Phototherapy may slightly increase the risk of cancer in infancy, although the absolute risk increase is small. This risk should be considered when making phototherapy treatment decisions, especially for infants with bilirubin levels below current treatment guidelines.

aDepartment of Pediatrics, Kaiser Permanente Northern California, Santa Clara, California; bDepartment of

Epidemiology & Biostatistics, and cDepartment of Pediatrics, University of California, San Francisco, California;

and dDivision of Research, Kaiser Permanente Northern California, Oakland, California

Dr Wickremasinghe assisted with study design, obtained funding, obtained data, carried out the

initial analyses, and drafted the initial manuscript; Dr Kuzniewicz assisted with study design and

reviewed and revised the manuscript; Dr Grimes provided statistical consultation and reviewed

and revised the manuscript; Dr McCulloch assisted with study design, provided statistical

consultation, and reviewed and revised the manuscript; Dr Newman conceptualized and designed

the study, assisted with obtaining funding and statistical analyses, and reviewed and revised the

manuscript; and all authors approved the fi nal manuscript as submitted.

DOI: 10.1542/peds.2015-1353

Accepted for publication Mar 25, 2016

Address correspondence to Andrea C. Wickremasinghe, MD, Department 302–Neonatology,

Kaiser Permanente Santa Clara Medical Center, 700 Lawrence Expy, Santa Clara, CA 95051. E-mail:

andrea.wickremasinghe@kp.org

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

To cite: Wickremasinghe AC, Kuzniewicz MW, Grimes BA, et

al. Neonatal Phototherapy and Infantile Cancer. Pediatrics.

2016;137(6):e20151353

WHAT’S KNOWN ON THIS SUBJECT: Phototherapy

is commonly used to treat neonatal jaundice. It

is generally thought to be safe. However, a small

number of studies have suggested an association

between neonatal phototherapy and cancer.

WHAT THIS STUDY ADDS: In this study of ~5 million

California births, phototherapy was associated with

a small but statistically signifi cant increased risk of

some infantile cancers, specifi cally myeloid leukemia

and kidney cancer, with ~1 additional cancer case

per 10 638 treated infants.

by guest on June 24, 2016Downloaded from

Department of Pediatrics, Kaiser Permanente Northern California, Santa Clara, California;

PEDIATRICS Volume 137 , number 6 , June 2016 :e 20151353

Experienced J-Clubbers know the significance of the order of authors: 1st actually writes the paper, last author is Senior who

oversaw the whole research effort. 2nd through penultimate are in decreasing order of contribution. Pro tip: when setting out to do

research, establish authors and orders BEFORE you start. You'll thank us later if you do.

We analyzed a data set from the California Office of Statewide Health Planning and

1st authoractuallywrites it.MD @ U of MO,peds @ Mayo,NICU @ UCSF.This young physicianhas 17 pubs as of 8/17

med school @ NWern

Andrea C. Wickremasinghe, MD, Michael W. Kuzniewicz, MD, MPH,

Thomas B. Newman, MD, MPH

Impact Factor 5.1

(how influential journal is)

(derived from how many times other journals cite itdivided by how many article journal publishes)

send her an email and tell her howthoroughly you enjoyed her writing

fun fact: many

former military

devo docs work

for Kaiser

WICKREMASINGHE et al

Phototherapy is commonly used to treat neonatal jaundice. The use of phototherapy has been increasing1, 2; a recent study of term and near-term infants enrolled in a single health plan in California showed that up to 15.9% received this treatment.3 Many clinicians initiate phototherapy at levels lower than those recommended by the American Academy of Pediatrics.2, 4 The rise in phototherapy use may be a result of increased identification of neonates with hyperbilirubinemia through universal screening protocols, 2 fear of kernicterus, 5 and general belief that phototherapy is safe.4Although phototherapy is thought to pose minimal risk to infants with unconjugated hyperbilirubinemia, 4, 6 2 large epidemiologic studies have found phototherapy to be associated with subsequent cancer (overall leukemia and acute myelogenous leukemia).7, 8 These associations have not been consistent; some studies have been reassuring, 9, 10 but none have been large enough to be conclusive. The potential for phototherapy to increase the incidence of cancer is supported by data from in vitro and in vivo studies.11–22

It is important to determine whether phototherapy contributes to early childhood cancer to determine the optimal management of infants with neonatal hyperbilirubinemia. In this study, we used population-based data to investigate the association between phototherapy and cancer in the first year after birth.

METHODS

Study Design and Subjects

This study is an analysis of a historical cohort of infants born in California between January 1, 1998 and December 31, 2007. We used the linked Vital Statistics/Patient Discharge Dataset (VS/PDD) for our analyses. The VS/PDD was created

for the California Office of Statewide Health Planning and Development by using probabilistic linkage of data from birth certificates, death certificates, and hospital discharge abstracts for mothers and their newborn infants up to 1 year of age.23 The VS/PDD includes all California births except those of infants born at home or in military hospitals. For our analyses, we excluded infants born at <35 weeks’ gestation and infants who died within 60 days of birth.This study was approved by the institutional review boards for the protection of human subjects at the California Health and Human Services Agency, California Department of Public Health, and the University of California, San Francisco.

Predictor Variables

The primary predictor variable was receipt of phototherapy during a hospitalization that began within the first 2 weeks after birth. We ascertained phototherapy use by abstracting International Classification of Diseases, Ninth Revision (ICD-9)24 procedure codes for phototherapy (99.82, 99.83).Covariates abstracted from the birth certificates included gender, birth weight, gestational age, twin birth, birth by cesarean delivery, payer source, year of birth, and maternal and paternal demographic variables (race, Hispanic ethnicity, age, and education). Covariates abstracted from discharge diagnosis ICD-9 codes included jaundice (774.0–774.4, 774.5, 774.6, 782.4), Down syndrome (758.0), other chromosomal anomalies (758.1–758.9), and nonchromosomal congenital anomalies (740.0–743.64, 743.66–749.25, 750.1–757.32, 757.39–757.9, 759.0–759.9); only congenital anomalies diagnosed during the birth hospitalization were included.Data were missing for several of our covariates, most commonly payer source (missing in 12.1%), paternal

education (9.4%), gestational age (7.1%), paternal age (7.1%), and paternal race (6.9%). We imputed missing values by using single regression imputation because of the large size of the data set. To estimate the effect of the missing data on the precision of our results, we performed multiple imputation (5 imputations) on a subset that included all cancer cases and a 1% sample of the remainder of the cohort. We included subjects with missing gestational age if their imputed gestational age (based on multiple variables including birth weight, parental race, parental age, mode of delivery, and twin status) was ≥35 weeks.Outcome Variables

The primary outcome was a hospital discharge diagnosis of cancer (ICD-9 codes 140.0–209.79, 230.0–234.9) in the first year after birth. We excluded cancers diagnosed at ≤60 days to reduce the likelihood of finding an association through reverse causation. In addition, it seemed less plausible that phototherapy would cause cancer in ≤60 days. We grouped cancer diagnosis ICD-9 codes by site: myeloid leukemia (205.0–205.92), lymphoid leukemia (204.0–204.92), brain or nervous system (191.0–192.9), kidney (189.0–189.9), liver (155.0–156.9), soft tissue (171.0–171.9), and other (remaining codes). If an infant had >1 type of cancer diagnosis, all were captured.Statistical Analysis

We used Stata 13 (Stata Corp, College Station, TX) statistical software for all analyses. We performed bivariate analyses comparing infants with and without cancer diagnoses by using χ2 tests for categorical variables and t tests for continuous variables.We computed risk ratios (RRs) and risk differences for phototherapy and each of the cancer outcomes by using the Agresti-Caffo method25 to

2by guest on June 24, 2016Downloaded from

that up to 15.9% received this 3 Many clinicians initiate treatment.3

???

and general belief that phototherapy is safe.4

2 large epidemiologic studies have found phototherapy to be associated with subsequent cancer (overall leukemia and acute myelogenous leukemia).7, 8

incidence of cancer is supported by data from in vitro and in vivo studies.11–22

data to investigate the association between phototherapy and cancer in the first year after birth.

Objectivebut

no

hypothesis

PICO

historical cohort of infants born in California between January 1, 1998 California between January 1, 1998 and December 31, 2007. We used the linked Vital Statistics/Patient Discharge Dataset (VS/PDD) for our

by using probabilistic linkage of data from birth certificates, death certificates, and hospital discharge abstracts for mothers and their newborn infants up to 1 year of age.

births except those of infants born at home or in military hospitals. For our analyses, we excluded infants born at <35 weeks’ gestation and infants who died within 60 days of birth.

what if born at civ hospital, readmitted to military?

procedure codes for phototherapy (99.82, 99.83).

these

are

vatiablres

thatw

illbe

'adju

ste

dfo

r'in

the

model-

als

oused

togenera

te

a 'p

ropensity

score

'

there are 2 ways to handle missing data: (1) throw the whole record

out unless EVERY variable is present, or (2) 'make up' a value for the

missing data point in a way that doesn't mess up the effect of the real data (aka IMPUTATION)

paternal race (6.9%). We imputed missing values by using single

An example of IMPUTATION in a group of 3 subjects

ID C/S Ethn Insurance Down Cancer

1 Y HISP PRIV N N

2 N ------ PUB N N

3 N nonHIS PUB N Y

If you want to adjust for ethnicity, then

subject 2 would be thrown out (along with

the valuable data that was collected). But

if you IMPUTE that subject 2 was half-

hispanic/half non-hispanic, mathematically

your made-up data wouldn't affect

anything.

skepticaljc

lub

readers

want

toknow

HO

Wim

puta

tion

was

done,

and

auth

ors

who

impute

bett

er

show

their

non-im

puta

tion

results

too.

The primary outcome was a hospital discharge diagnosis of cancer (ICD-9

the first year after birth. We excluded ≤cancers diagnosed at ≤60 days to

We grouped cancer diagnosis ICD-9 codes by site: myeloid leukemia

expect multiple

analyses: one

for all CA and

one for each

subgroup

We used Stata 13 (Stata Corp, College

the other

prof stats

software are

SAS, SPSS,

and R.

NCCPeds

has 13

licenses for

Stata for

@NCCPeds

Res to use.

R is open

source

without cancer diagnoses by using χ2without cancer diagnoses by using ttests for categorical variables and

tests for continuous variables.'bread-n-butter' stats, although what adjective should precede 'continous'?

Risk Ratios are

calculated from

2x2 tables; they

can't be

modeled in

regressions so

surrogates like

Odds Ratios,

Relative Rates

are used to

approximate.

We computed risk ratios (RRs) and

RD = RR - RR. A-C method

calculates the CI. RD will be

used to generate NNT or NNH

as (1/RD)

PEDIATRICS Volume 137 , number 6 , June 2016

compare risk differences. Because the exposure (phototherapy) is much more common than the outcome (various types of cancer), creating a model for exposure allowed control for many more potential confounding variables than traditional analyses that create a model for the outcome. We therefore created a propensity score for phototherapy with the following covariates: gender, birth weight, gestational age, large for gestational age, twin birth, birth by cesarean delivery, payer source, year of birth, maternal race, paternal race, maternal Hispanic ethnicity, maternal age, paternal age, maternal education, paternal education, Down syndrome, other chromosomal anomalies, and nonchromosomal congenital anomalies. The propensity score had an area under the receiver operating characteristic curve of 0.67. We then used logistic regression to evaluate the association between phototherapy and cancer, adjusted for the propensity to receive phototherapy. In the propensity-adjusted analyses, the predictor variables were only phototherapy and propensity to receive phototherapy (in categories). We calculated unadjusted and marginal propensity-adjusted risk differences and 95% confidence intervals (CIs) of the risk differences; we calculated marginal risk differences because they estimate effect in the entire population (ie, if all versus no infants were treated with phototherapy), whereas other risk differences depend on baseline risk (ie, the values of other covariates).26 We calculated numbers needed to harm by taking the reciprocal of the risk differences. To complement propensity-adjusted models, we also analyzed the data in traditional multivariable logistic models adjusted for gender, birth weight, gestational age, Down syndrome, maternal race, and year of birth. To ensure that imputation did not affect our results, we performed a complete case analysis.

Because Down syndrome is such a strong risk factor for cancer, 27 we estimated the marginal propensity-adjusted risk difference and 95% CI for cancer after phototherapy in infants with and without Down syndrome separately. For this analysis, we created a propensity score without the Down syndrome variable.

RESULTS

Our cohort consisted of 5 144 849 infants born at ≥35 weeks’ gestation who were alive >60 days after birth. A cancer diagnosis was present

in 1100 of the infants, yielding an overall population incidence of 21.4/100 000. Table 1 provides a description of the cohort by cancer diagnosis status. The most common cancers were leukemia (23%), brain or nervous system cancer (16%), and eye or orbit cancer (9%).Jaundice diagnoses were recorded in 13.9% of infants and phototherapy codes in 3.5% of the cohort. Phototherapy use increased over time, from 2.9% in 1998 to 4.4% in 2007 (P < .001). Of subjects who received phototherapy, 85% had a jaundice diagnosis. Jaundice itself was associated with an increased

3

TABLE 1 Description of Cohort (n = 5 144 861), by Diagnosis of Cancer

Cancer Diagnosis P

Yes (n = 1100) No (n = 5 143 749)

Male, % 55.2 51 .005

Gestation in wk, mean (SD) 39.5 (1.7) 39.5 (1.6) .65

Birth wt in g, mean (SD) 3411 (511) 3385 (503) .09

Large for gestational age, %a 10.9 9.9 .28

Twin, % 2.4 2.2 .74

Down syndrome, % 1.6 0.2 <.001

Other chromosomal anomalies, % 0.6 0.1 <.001

Congenital anomalies, % 7.7 4.1 <.001

Maternal race, % .1

White 83 80.2

Black 5 5.8

Asian 10.2 11.4

Other 1.9 2.6

Maternal Hispanic ethnicity, % 50.6 47.5 .04

Paternal race, % .36

White 82.4 80.2

Black 5.8 6.7

Asian 9.4 10.6

Other 2.4 2.6

Maternal age in y, mean (SD) 28.2 (6.4) 27.9 (6.3) .08

Paternal age in y, mean (SD) 30.9 (7) 30.8 (7.1) .58

Maternal education, % .78

Did not complete high school 31.4 29.8

Completed high school, no college 26.3 27.4

Completed some college 20.2 20

Completed 4 y of college 12.6 13.1

Completed >4 y of college 9.5 9.8

Paternal education, % .7

Did not complete high school 30.4 28.5

Completed high school, no college 29.2 29.2

Completed some college 16.7 17.9

Completed 4 y of college 12.8 13.2

Completed >4 y of college 11 11.1

Cesarean delivery, % 29.5 26.2 .02

Payer, % .07

Private insurance 50.6 51.5

Medi-Cal 46.7 44.6

Other 2.7 4

a Large for gestational age is defi ned as >90th percentile (in this data set).

by guest on June 24, 2016Downloaded from

Purists should be suspicious if Table 1 doesn't show the demographics of the groups. This is a true Table 1, but separates

groups not by the exposure (photoTx) but by outcome. Weird.

We therefore created a propensity score for phototherapy with the following covariates: gender, birth

sin

ce

retr

ospective

stu

die

slik

eth

iscannot

be

random

ized,

the

pro

pensity

score

att

em

pts

tow

eig

ht

the

contr

ibution

ofeach

subje

ct

based

on

how

likely

they

were

togetphoto

Tx.

They

do

are

gre

ssio

nofth

eoutc

om

eP

hoto

Tx

based

on

the

covariate

s,

then

assig

neach

subje

ct

aS

CO

RE

.T

he

SC

OR

Eth

en

becom

es

acovariate

alo

ng

with

all

the

oth

ers

inth

eC

AN

CE

Ranaly

sis

.

Ask

yours

elf:

why

would

babie

sw

ith

Xbe

more

likely

toget

Cancer?

Ifth

ere

'sno

good

possib

lere

ason,

and

gett

ing

photo

thera

py

isn't

connecte

d,

maybe

there

'sanoth

er

reason

thatis

n't

bein

gaccounte

dfo

r.C

an

you

hypoth

esiz

ea

reason

for

each

ofth

ese

red

boxes.

congenital anomalies. The propensity score had an area under the receiver operating characteristic curve of 0.67. We then used logistic

LO

GIS

TIC

->O

dds

Ratios

PO

ISS

ON

->R

ate

Ratios

CO

X->

Hazard

Ratio

big #

also a big #

jaundice diagnosis. Jaundice itself was associated with an increased

this is unadjusted

false+

true+

ROC isthis plotof how wellprop scorefits data.the # is thearea under it

0.67

calculated unadjusted and marginal propensity-adjusted risk differences and 95% confidence intervals (CIs)

marginal risk differences because they estimate effect in the entire population (ie, if all versus no infants were treated with phototherapy),

We calculated numbers needed to harm by taking the reciprocal of the risk differences. To complement

also analyzed the data in traditional multivariable logistic models adjusted for gender, birth weight, adjusted for gender, birth weight, adjusted for gender, birth weight, adjusted for gender, birth weight, gestational age, Down syndrome, maternal race, and year of birth. To gestational age, Down syndrome, gestational age, Down syndrome, maternal race, and year of birth. To

good authors! don't imputewithout showing us what justthe real stuff shows

separate analysis

WICKREMASINGHE et al

risk of cancer, controlling for phototherapy (RR 1.3; 95% CI, 1.1–1.5).

In unadjusted analyses, subjects who received phototherapy were at higher risk of overall cancer (RR 1.6; 95% CI, 1.2–2.0), myeloid leukemia (RR 2.7; 95% CI, 1.4–5.2), kidney cancer (RR 2.7; 95% CI, 1.3–5.6), and other cancer (RR 1.7; 95% CI, 1.1–2.5), although the risk differences were small (Tables 2 and 3). The associations between phototherapy and overall cancer, myeloid leukemia, kidney cancer, and other cancer persisted in propensity-adjusted models (Table 3). Propensity-adjusted models, traditional multivariable-adjusted models, and models with and without multiple imputation gave similar results. In a complete case analysis (n = 3 717 503), we found similar propensity-adjusted associations between phototherapy and overall cancer (odds ratio [OR] 1.5; 95% CI, 1.1–2.1), myeloid leukemia (OR 2.2; 95% CI, 1.0–4.9), kidney cancer (OR 3.4; 95% CI, 1.6–7.3), and other cancer (OR 1.7; 95% CI, 1.1–2.7). After adjustment for jaundice alone, the association between phototherapy and myeloid leukemia was similar (OR 2.6; 95% CI, 1.2–5.6), the associations between phototherapy and overall cancer (OR 1.3; 95% CI, 1.0–1.8) and other cancer (OR 1.5; 95% CI, 1.0–2.4) were slightly diminished, and the association between phototherapy and kidney cancer (OR 1.6; 95% CI, 0.7–3.5) was diminished further. The marginal propensity-adjusted absolute risk increase for cancer after phototherapy in the total population was 9.4/100 000 (95% CI, 1.4/100 000–17.4/100 000), with a number needed to harm of 10 638.Down syndrome was diagnosed in 7812 subjects (0.2% of the cohort). In the subgroup of infants with Down syndrome, 19% received phototherapy. Cancer was diagnosed

in 18 infants with Down syndrome (lymphoid leukemia in 4, myeloid leukemia in 10, other leukemia in 2, brain or nervous system cancer in 1, and other cancer in 1). The marginal propensity-adjusted absolute risk increase for cancer after phototherapy was much higher for infants with Down syndrome (77.8/100 000; 95%

CI, -1.2/100 000–156.8/100 000, number needed to harm of 1285) than for infants without Down syndrome (8.1/100 000, 95% CI 0.4/100 000–15.8/100 000, number needed to harm of 12 346). This difference was caused by the higher baseline risk of cancer in infants with Down syndrome; the propensity-adjusted ORs for cancer after

4

TABLE 2 Incidence of Cancer by Whether Infants Received Phototherapy, by Cancer Site

Phototherapy (n

= 178 017)

No Phototherapy

(n = 4 966 832)

Unadjusted

Risk

Difference

(per

100 000)

95% CI

(per

100 000)

Pa

Nb Incidence

(per

100 000)

Nb Incidence

(per

100 000)

All cancer 58 32.6 1042 21.0 11.6 3.6 to

20.7

.002

Leukemia 16 9.0 242 4.9 4.1 0.1 to 9.2 .025

Lymphoid leukemia 6 3.4 133 2.7 0.7 −1.7 to

4.2

.49

Myeloid leukemia 10 5.6 103 2.1 3.5 0.4 to 7.8 .006

Brain or nervous system

cancer

7 3.9 176 3.5 0.4 −2.2 to

4.1

.69

Eye or orbit cancer 4 2.3 103 2.1 0.2 −1.8 to

3.2

.79

Kidney cancer 8 4.5 82 1.7 2.8 0.1 to 6.7 .013

Liver cancer 2 1.1 81 1.6 −0.5 −1.9 to

2.0

1.0

Soft tissue cancer 1 0.6 69 1.4 −0.8 −1.9 to

1.3

.52

Other cancer 26 14.6 426 8.6 6.0 0.8 to

12.4

.013

a Fisher’s exact test used.b Some subjects had >1 type of cancer.

TABLE 3 RRs and Propensity-Adjusted ORs for the Association Between Phototherapy and Cancer, by

Cancer Site

Na Unadjusted Analyses Propensity-Adjusted Analysesb

RR 95% CI Pc OR 95% CI P

All cancer 1100 1.6 1.2 to 2.0 .002 1.4 1.1 to 1.9 .007

Leukemia 258 1.8 1.1 to 3.1 .03 1.8 1.1 to 3.1 .02

Lymphoid leukemia 139 1.3 0.6 to 2.9 .49 1.3 0.6 to 2.9 .55

Myeloid leukemia 113 2.7 1.4 to 5.2 .006 2.6 1.3 to 5.0 .005

Brain or nervous system

cancer

183 1.1 0.5 to 2.4 .69 1.0 0.5 to 2.1 .97

Eye or orbit cancer 107 1.1 0.4 to 2.9 .79 1.0 0.4 to 2.7 .96

Kidney cancer 90 2.7 1.3 to 5.6 .01 2.5 1.2 to 5.1 .02

Liver cancer 83 0.7 0.2 to 2.8 1.0 0.6 0.2 to 2.5 .5

Soft tissue cancer 70 0.4 0.1 to 2.9 .52 0.4 0.1 to 2.7 .33

Other cancer 452 1.7 1.1 to 2.5 .01 1.6 1.1 to 2.4 .02

a Some subjects had >1 type of cancer.b Propensity-adjusted model included a propensity score for phototherapy that included the following variables: gender,

birth wt, gestational age, large for gestational age, twin birth, birth by cesarean delivery, payer source, year of birth,

maternal race, paternal race, maternal Hispanic ethnicity, maternal age, paternal age, maternal education, paternal

education, Down syndrome, other chromosomal anomalies, and nonchromosomal congenital anomalies. The propensity

score had an area under the receiver operating characteristic curve of 0.67.c Fisher’s exact test used.

by guest on June 24, 2016Downloaded from

risk of cancer, controlling for phototherapy (RR 1.3; 95% CI, 1.1–1.5).

this is unadjusted

purists don't need p-value if 95%

confidence interval provided. RR sig if CI doesnt cross 1

(RR 1.6; 95% CI, 1.2–2.0), myeloid leukemia (RR 2.7; 95% CI, 1.4–5.2), kidney cancer (RR 2.7; 95% CI, 1.3–5.6), and other cancer (RR 1.7; 1.3–5.6), and other cancer (RR 1.7; 95% CI, 1.1–2.5), although the risk

none

cross

1

3). Propensity-adjusted models, traditional multivariable-adjusted models, and models with and without models, and models with and without multiple imputation gave similar

models

agree

noticethey

switchedto

Odds

Ratiosfrom

RR

(RR

cannotbe

modeledby

regression,but

ORs

approximateRR

if

theoutcome

israre,

andcan

be

modeledif

eachterm

raisedto

the

magic

number'e'

aka

2.713

...)

cancer (odds ratio [OR] 1.5; 95% CI, 1.1–2.1), myeloid leukemia

sig if

doesnt

cross

1

CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for CI, 1.1–2.7). After adjustment for jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association jaundice alone, the association

(OR 1.3; 95% CI, 1.0–1.8) and other

The marginal propensity-adjusted absolute risk increase for cancer after phototherapy in the total

number needed to harm of 10 638.NNH = 1 / (difference in risk)

try it

out on

your

phone

calculator

calculated with Agresti-Caffo method

can u explain why RR on left and OR on right?

text Dr G to check your answer

Fisher’s exact test used. Red Alert! If you ever see 'Fischer Exact' it means some box of a 2x2table had a teeny # in it so this lesser form of stat test had to be used.Mention that matter-of-factly the next time you see it in a paper.

PEDIATRICS Volume 137 , number 6 , June 2016

phototherapy were similar: 1.2 for infants with Down syndrome and 1.4 for infants without Down syndrome (P for interaction = .79).

DISCUSSION

In this study, we found that neonatal phototherapy was associated with an increased risk of cancer in the first year after birth; these associations persisted in adjusted analyses. Specifically, we found associations between phototherapy and overall cancer (adjusted odds ratio [aOR] 1.4), myeloid leukemia (aOR 2.6), kidney cancer (aOR 2.5), and other cancer (aOR 1.6). Based on our findings, 10 638 infants would need to be treated with neonatal phototherapy to cause 1 excess case of infantile cancer.With epidemiologic studies, it is important to consider whether associations found are biologically plausible. The potential for phototherapy to increase the incidence of cancer is supported by evidence dating from the 1970s that blue light is mutagenic in vitro.11 Additionally, in vivo experiments on human newborns have demonstrated DNA damage, alterations in cytokine levels, and evidence of oxidative stress after treatment with phototherapy.12–22 This is concerning because all these conditions have been implicated in the pathogenesis of cancer.28, 29 Although the clinical importance of these alterations remains to be elucidated, they present a potential mechanism whereby phototherapy could be related to childhood cancer.The association between phototherapy and leukemia reported here has been previously suggested in other large epidemiologic studies. A population-based case–control study from the state of Washington found neonatal jaundice to be associated with an increased risk of leukemia (OR 1.5; 95% CI, 1.1–2.1).7In a subgroup analysis, the risk of

leukemia was higher in jaundiced infants who received phototherapy (OR 2.2; 95% CI, 1.0–4.9).7 A Swedish case–control study found increased odds of myeloid leukemia in infants with physiologic jaundice (OR 2.5; 95% CI, 1.2–5), which again was more pronounced in those treated with phototherapy (OR 7.5; 95% CI, 1.8–31.9).8 As in the current study, the Swedish group did not find an association between phototherapy and lymphoid leukemia (OR 1.0; 95% CI, 0.5–1.8).30

Two other studies did not find an association between phototherapy and cancer. A case–control study from the United Kingdom with 143 leukemia cases found no statistically significant association between phototherapy and leukemia (OR 0.5; 95% CI, 0.1–2.3); in this study, none of the 15 subjects with myeloid leukemia received phototherapy (OR 0; 95% CI, 0–11.7).10 A Danish study of 55 120 children with neonatal hyperbilirubinemia found no association between hyperbilirubinemia and overall cancer (standardized incidence ratio [SIR] 1.0; 95% CI, 0.8–1.3) or any subtype, including leukemia (SIR 1.2; 95% CI, 0.8–1.7) and kidney cancer (SIR 1.0; 95% CI, 0.4–2.2).9 The authors of that study estimated that 85% to 90% of the children with hyperbilirubinemia received phototherapy, but they were unable to specifically evaluate the association between phototherapy and cancer.In this study, we also found an association between phototherapy and other cancer. The “other cancer” group included an excess of bone cancer, which is extremely rare in infancy, suggesting that this group consists, at least in part, of metastases and misclassified cancers. Nonetheless, we found a persistent association between phototherapy and these other cancers.Our study has the advantage of a large cohort, with >5 million infants

and 1100 cancer cases. The incidence of cancer in our study is similar to what is reported from US Cancer Registries, 31 which suggests that our ascertainment of cancer cases was reasonably complete.Because we used a large statewide data set, this study has a few limitations. We were unable to adjust for potential confounding variables not included in the data set. As an example, because serum bilirubin levels were not available, we cannot exclude the possibility that the association between phototherapy and cancer could be caused by elevated bilirubin levels (not captured by a discharge diagnosis of jaundice). The association between phototherapy and cancer persisted in analyses adjusted for a discharge diagnosis of jaundice, although a diagnosis of jaundice is admittedly a crude indicator of hyperbilirubinemia.Because data on phototherapy intensity and duration were not available, we could not investigate the effect of phototherapy dosage. We did not have data on home phototherapy; we hypothesize that misclassification of subjects who received home phototherapy as unexposed would attenuate the observed association between phototherapy and cancer. Our ascertainment of phototherapy and of cancer diagnoses was probably incomplete, but it is unlikely that the completeness or accuracy of cancer coding differed between those who did and did not receive phototherapy or that the accuracy of phototherapy coding differed depending on subsequent risk of cancer. Such nondifferential misclassification, if present, would have biased our ORs toward 1. Also, the data set we used is limited to the first year after birth.In a companion study, Newman et al3 looked at the association between neonatal phototherapy and childhood cancer in a cohort of almost 500 000 children born over a

5by guest on June 24, 2016Downloaded from

phototherapy was associated with an

these kinds ofstudies don'tshow causation

findings, 10 638 infants would need to be treated with neonatal phototherapy to cause 1 excess case of infantile cancer.

afew

paragraphsonconsistency

with

previousstudies

and other cancer. The “other cancer” group included an excess of bone cancer, which is extremely rare

metastases and misclassified cancers.

ascertainment of cancer cases was reasonably complete.

limitations. We were unable to adjust limitations. We were unable to adjust for potential confounding variables

Because data on phototherapy intensity and duration were not available, we could not investigate

We did not have data on home phototherapy; we hypothesize

of cancer diagnoses was probably incomplete, but it is unlikely that the

wait, does'reasonablycomplete'mean the sameas ' probincomplete'?

nondifferential misclassification, if

that misclassification of subjects

misclassification

meansputting

kids

inwrong

category.If

it

does

it

equallyboth

ways,then

its

non-differential.

that's me

again!

In a companion study, Newman

WICKREMASINGHE et al

17-year period in Kaiser Permanente Northern California hospitals. In that study, phototherapy was associated with an increased risk of overall cancer and nonlymphocytic leukemia in unadjusted models, but these associations were no longer statistically significant after researchers adjusted for potential confounders. These results could reflect the ability to adjust for additional potential confounders that were not available in our data set (eg, bilirubin levels). However, the 95% CI reported in that study did not exclude adjusted differences of the magnitude reported here. It is also possible that phototherapy causes only infantile or early childhood cancer; the Newman et al finding of increased early (<3 years) nonlymphocytic leukemia could support this hypothesis.Although we found an increased risk of cancer in infants who received phototherapy, the absolute risk increase among those who received phototherapy was low. However, for infants with Down syndrome, the absolute risk increase was almost 10 times higher because of their higher baseline risk for cancer.The possible increase in cancer risk from using phototherapy must

be balanced against its benefit for lowering bilirubin levels. Phototherapy use has decreased rates of exchange transfusions, 32 a procedure estimated to have 5% morbidity and up to 1.9% mortality risks.33 However, the number of infants with bilirubin levels near phototherapy treatment thresholds who need to be treated to prevent 1 from reaching exchange transfusion thresholds varies widely (from 10 to >3000, based on gender, gestational age, and age).34 Additionally, having bilirubin levels at or just above exchange transfusion thresholds is typically benign; it is not until bilirubin levels are >5 to 10 mg/dL above exchange transfusion thresholds that the risks of cerebral palsy and sensorineural hearing loss increase significantly.35–40 Therefore, especially in infants with bilirubin levels below recommended treatment thresholds and in infants with Down syndrome, the risks of phototherapy may exceed the benefits.

CONCLUSIONS

Phototherapy is a valuable treatment for infants with neonatal hyperbilirubinemia. Our results support the need for health care providers to consider phototherapy

as they do most other treatments (ie, as having both benefits and potential risks) and to limit the use of phototherapy to infants for whom the benefit/risk balance is most likely to be favorable.

ACKNOWLEDGMENTS

The authors thank Dr Beate Danielsen for her assistance on this project. Statistical analysis for this project was partially supported by the National Center for Advancing Translational Sciences, National Institutes of Health, through UCSF-CTSI grant UL1 TR000004. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.

REFERENCES

1. Burke BL, Robbins JM, Bird TM, Hobbs

CA, Nesmith C, Tilford JM. Trends

in hospitalizations for neonatal

jaundice and kernicterus in the

United States, 1988–2005. Pediatrics.

2009;123(2):524–532

2. Kuzniewicz MW, Escobar GJ,

Newman TB. Impact of universal

bilirubin screening on severe

hyperbilirubinemia and phototherapy

use. Pediatrics. 2009;124(4):

1031–1039

3. Newman TB, Wickremasinghe AC,

Walsh EM, Grimes BA, McCulloch CE,

Kuzniewicz MW. Retrospective cohort

study of phototherapy and childhood

cancer in northern California.

Pediatrics. 2016;137(6):e20151334

4. American Academy of Pediatrics

Subcommittee on Hyperbilirubinemia.

Management of hyperbilirubinemia

in the newborn infant 35 or more

weeks of gestation. Pediatrics.

2004;114(1):297–316

5. Davidson L, Thilo EH. How to make

kernicterus a “never event.”

NeoReviews. 2003;4(11):e308–e314

6

ABBREVIATIONS

aOR:  adjusted odds ratioCI:  confidence intervalICD-9:  International

Classification of Diseases, Ninth Revision

OR:  odds ratioRR:  risk ratioSIR:  standardized incidence ratioVS/PDD:  Vital Statistics/Patient

Discharge Dataset

Copyright © 2016 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they have no fi nancial relationships relevant to this article to disclose.

FUNDING: Gerber Foundation Novice Research Award and American Academy of Pediatrics Marshall Klaus Perinatal Research Award to A.C.W.

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential confl icts of interest to disclose.

Companions to this article can be found online at www. pediatrics. org/ cgi/ doi/ 10. 1542/ peds. 2015- 1354 and www. pediatrics. org/ cgi/ doi/ 10. 1542/ peds. 2016- 0983.

by guest on June 24, 2016Downloaded from

Although we found an increased risk of cancer in infants who received phototherapy, the absolute risk increase among those who received phototherapy was low.

Therefore, especially in infants with bilirubin levels below recommended treatment thresholds and in infants with Down syndrome, the risks of phototherapy may exceed the benefits.

The authors have indicated they have no fi nancial relationships relevant to this article to disclose.

Gerber Foundation Novice Research Award and American Academy of Pediatrics Marshall Klaus Perinatal Research Award to A.C.W. Gerber Foundation Novice Research Award and American Academy of Pediatrics Marshall Klaus Perinatal Research Award to A.C.W.

look at this section

carefully before

reading article

The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose. The authors have indicated they have no potential confl icts of interest to disclose.

PEDIATRICS Volume 137 , number 6 , June 2016

6. Maisels MJ, McDonagh AF.

Phototherapy for neonatal jaundice.

N Engl J Med. 2008;358(9):920–928

7. Podvin D, Kuehn CM, Mueller BA,

Williams M. Maternal and birth

characteristics in relation to childhood

leukaemia. Paediatr Perinat Epidemiol.

2006;20(4):312–322

8. Cnattingius S, Zack M, Ekbom A,

Gunnarskog J, Linet M, Adami HO.

Prenatal and neonatal risk factors

for childhood myeloid leukemia.

Cancer Epidemiol Biomarkers Prev.

1995;4(5):441–445

9. Olsen JH, Hertz H, Kjaer SK, Bautz A,

Mellemkjaer L, Boice JD Jr. Childhood

leukemia following phototherapy

for neonatal hyperbilirubinemia

(Denmark). Cancer Causes Control.

1996;7(4):411–414

10. Roman E, Ansell P, Bull D. Leukaemia

and non-Hodgkin’s lymphoma in

children and young adults: are

prenatal and neonatal factors

important determinants of disease? Br

J Cancer. 1997;76(3):406–415

11. Speck WT, Rosenkranz HS.

Phototherapy for neonatal

hyperbilirubinemia—a potential

environmental health hazard to

newborn infants: a review. Environ

Mutagen. 1979;1(4):321–336

12. Tsai FJ, Tsai CH, Peng CT, Wang

TR. Sister chromatid exchange in

Chinese newborn infants treated with

phototherapy for more than fi ve days.

Zhonghua Minguo Xiao Er Ke Yi Xue

Hui. 1998;39(5):327–329

13. Galla A, Kitsiou-Tzeli S, Gourgiotis D,

et al. Sister chromatid exchanges in

peripheral lymphocytes in newborns

treated with phototherapy and vitamin

E. Acta Paediatr. 1992;81(10):820–823

14. Karadag A, Yesilyurt A, Unal S,

et al. A chromosomal-effect study

of intensive phototherapy versus

conventional phototherapy in

newborns with jaundice. Mutat Res.

2009;676(1-2):17–20

15. Tatli MM, Minnet C, Kocyigit A,

Karadag A. Phototherapy increases

DNA damage in lymphocytes of

hyperbilirubinemic neonates. Mutat

Res. 2008;654(1):93–95

16. Aycicek A, Erel O. Total oxidant/

antioxidant status in jaundiced

newborns before and after

phototherapy. J Pediatr (Rio J).

2007;83(4):319–322

17. Aycicek A, Kocyigit A, Erel O, Senturk

H. Phototherapy causes DNA damage

in peripheral mononuclear leukocytes

in term infants. J Pediatr (Rio J).

2008;84(2):141–146

18. Gathwala G, Sharma S. Oxidative

stress, phototherapy and the neonate.

Indian J Pediatr. 2000;67(11):805–808

19. Gathwala G, Sharma S. Phototherapy

induces oxidative stress in premature

neonates. Indian J Gastroenterol.

2002;21(4):153–154

20. Sirota L, Straussberg R, Gurary N,

Aloni D, Bessler H. Phototherapy for

neonatal hyperbilirubinemia affects

cytokine production by peripheral

blood mononuclear cells. Eur J Pediatr.

1999;158(11):910–913

21. Kurt A, Aygun AD, Kurt AN,

Godekmerdan A, Akarsu S, Yilmaz

E. Use of phototherapy for neonatal

hyperbilirubinemia affects cytokine

production and lymphocyte subsets.

Neonatology. 2009;95(3):262–266

22. Ramy N, Ghany EA, Alsharany W, et

al Jaundice, phototherapy and DNA

damage in full-term neonates. J

Perinatol. 2016;36(2):132–136

23. Herrchen B, Gould JB, Nesbitt TS. Vital

statistics linked birth/infant death

and hospital discharge record linkage

for epidemiological studies. Comput

Biomed Res. 1997;30(4):290–305

24. Practice Management Information

Corporation. ICD-9-CM: International

Classifi cation of Diseases, 9th Revision,

Clinical Modifi cation, Fifth Edition,

Color Coded, 2000. Hospital Edition.

Los Angeles, CA: Practice Management

Information Corporation; 1999

25. Agresti A, Caffo B. Simple and effective

confi dence intervals for proportions

and differences of proportions result

from adding two successes and two

failures. Am Stat. 2000;54(4):280–288

26. Williams R. Using the margins

command to estimate and interpret

adjusted predictions and marginal

effects. Stata J. 2012;12(2):308–331

27. Fong CT, Brodeur GM. Down’s syndrome

and leukemia: epidemiology, genetics,

cytogenetics and mechanisms of

leukemogenesis. Cancer Genet

Cytogenet. 1987;28(1):55–76

28. Dranoff G. Cytokines in cancer

pathogenesis and cancer therapy. Nat

Rev Cancer. 2004;4(1):11–22

29. Valko M, Rhodes CJ, Moncol J, Izakovic

M, Mazur M. Free radicals, metals

and antioxidants in oxidative stress-

induced cancer. Chem Biol Interact.

2006;160(1):1–40

30. Cnattingius S, Zack MM, Ekbom A, et al.

Prenatal and neonatal risk factors for

childhood lymphatic leukemia. J Natl

Cancer Inst. 1995;87(12):908–914

31. US Cancer Statistics. 1999–2011

Incidence. WONDER online database.

2014. Available at: http:// wonder. cdc.

gov/ cancer- v2011. html. Accessed

February 5, 2015

32. Steiner LA, Bizzarro MJ, Ehrenkranz

RA, Gallagher PG. A decline in the

frequency of neonatal exchange

transfusions and its effect on

exchange-related morbidity

and mortality. Pediatrics.

2007;120(1):27–32

33. Ip S, Chung M, Kulig J, et al;

American Academy of Pediatrics

Subcommittee on Hyperbilirubinemia.

An evidence-based review of

important issues concerning neonatal

hyperbilirubinemia. Pediatrics.

2004;114(1). Available at: www.

pediatrics. org/ cgi/ content/ full/ 114/ 1/

e130

34. Newman TB, Kuzniewicz MW, Liljestrand

P, Wi S, McCulloch C, Escobar GJ.

Numbers needed to treat with

phototherapy according to American

Academy of Pediatrics guidelines.

Pediatrics. 2009;123(5):1352–1359

35. Wu YW, Kuzniewicz MW,

Wickremasinghe AC, et al. Risk for

cerebral palsy in infants with total

serum bilirubin levels at or above the

exchange transfusion threshold: a

population-based study. JAMA Pediatr.

2015;169(3):239–246

36. Wickremasinghe AC, Risley RJ,

Kuzniewicz MW, et al. Risk of

sensorineural hearing loss and

bilirubin exchange transfusion

thresholds. Pediatrics.

2015;136(3):505–512

37. Kuzniewicz MW, Wickremasinghe

AC, Wu YW, et al. Incidence, etiology,

7by guest on June 24, 2016Downloaded from

Authors have a 'G-index' that is related to the number of times they are cited (specificallya set of papers has a g-index g if g is the highest rank such that the top g papers have, together,at least g^2 citations. Citing yourself 6 times is like a self-licking ice-cream cone and is knownas academic self-stimulation.

WICKREMASINGHE et al

and outcomes of hazardous

hyperbilirubinemia in newborns.

Pediatrics. 2014;134(3):504–509

38. Ebbesen F, Bjerre JV, Vandborg

PK. Relation between serum

bilirubin levels ≥450 μmol/L and

bilirubin encephalopathy; a Danish

population-based study. Acta Paediatr.

2012;101(4):384–389

39. Vandborg PK, Hansen BM, Greisen G,

Jepsen M, Ebbesen F. Follow-up of neo-

nates with total serum bilirubin levels

≥25 mg/dL: a Danish population-based

study. Pediatrics. 2012;130(1):61–66

40. Vandborg PK, Hansen BM, Greisen

G, Mathiasen R, Kasper F, Ebbesen

F. Follow-up of extreme neonatal

hyperbilirubinaemia in 5- to 10-year-

old children: a Danish population-

based study. Dev Med Child Neurol.

2015;57(4):378–384

8by guest on June 24, 2016Downloaded from

dr Newman gettin' an award